[Intel-gfx] [PATCH] tests: debugfs_psr_status

[Rodrigo Vivi]

On Tue, Oct 1, 2013 at 5:31 PM, Daniel Vetter <daniel at ffwll.ch> wrote:
> On Tue, Oct 1, 2013 at 10:25 PM, Rodrigo Vivi <rodrigo.vivi at gmail.com> wrote:
>> On Tue, Oct 1, 2013 at 4:17 PM, Daniel Vetter <daniel at ffwll.ch> wrote:
>>> On Tue, Oct 01, 2013 at 03:26:43PM -0300, Rodrigo Vivi wrote:
>>>> Check on debugfs if PSR is supported by panel and matching all conditions in
>>>> hardware. In this case PSR must be enabled and performance counting increasing
>>>>
>>>> Signed-off-by: Rodrigo Vivi <rodrigo.vivi at gmail.com>
>>>
>>> Does this work if the console is blanked or someone just never enabled the
>>> screen? I.e. I think we need the usual boilerplate here to make sure that
>>> the eDP port is lit up on pipe A and dpms is ON ...
>>
>> in any case source doesn't match all conditions to get PSR enabled,
>> igt test will skip on this:
>>  if (!source)
>
> Ah, I see. Still the test is a bit fragile. E.g. when the bios ever
> enables PSR we won't fail the test, even when we don't enable psr. All
> our other tests that check residency of power save features prepare
> the system (make sure the gpu is idle for rc6, switch off all displays
> for pc8+, ...). Then they grab the residency counter, sleep for a few
> seconds and grab another sample of the residency counter. If it didn't
> move (or in case of rc6 we expect 90%+ residency even) then the test

got your point...
I was also wondering something like that... wait a bit and get perf
counter again... then it fails if it isn't increasing...
will do that

but I more concerned right now with the fix itself otherwise psr will
never work with 3.12 :(


> fails.
> -Daniel
> --
> Daniel Vetter
> Software Engineer, Intel Corporation
> +41 (0) 79 365 57 48 - http://blog.ffwll.ch



-- 
Rodrigo Vivi
Blog: http://blog.vivi.eng.br